On Fri, Dec 21, 2007 at 12:36:44 +0000, Paul Metcalfe wrote: > ... > > To the OP: the quick solution is to build your own atlas, lapack & > numpy. :-( > > ...
For the record, the NumPy bug tracker has a really quick solution: copy the matrix after loading it and before multiplying with it. import numpy phi = numpy.load('phi.p') phi = phi.copy() # This avoids the segfault! gamma = numpy.ones(17, dtype = float) out = numpy.dot(phi, gamma) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]